Reconstructing the investment method of early community projects? Gitcoin, the keyword worth tens of millions in DeFi

Editor's Note: This article comes fromDecentralized financial community (ID: China_DeFi), reprinted by Odaily with authorization.

Editor's Note: This article comes from

Decentralized financial community (ID: China_DeFi)

Decentralized financial community (ID: China_DeFi)

• , reprinted by Odaily with authorization.

• What is Gitcoin? It is a community product of an algorithm concept proposed by Vitalik to solve the problem of low financing efficiency of Ethereum ecological projects. Another more famous algorithm concept realized by the community is Uniswap.

• If Uniswap reconstructed the trading experience of the secondary market through the AMM algorithm on ETH1.0, then Gitcoin reconstructed the seed investment method for early community projects through the "second-order voting" algorithm. Notable projects funded through Gitcoin include: Tornado.cash, Uniswap, 1inch.exchange, Dune Analytics, MolochDAO

Sources of information valuable to investors under the keyword Gitcoin include:

Each issue of the Gitcoin Grand list is invested with real money, and you can find them when the projects are not the hottest. If you go back to the list of issues 1 to 8, you will find that the well-known DeFi projects have been in the appeared here.

Vitalik's summary of each issue of gitcoin grand is a very good directional information, which can obtain framework and real-time industry analysis, which can be viewed as a quarterly report.

If you still want to see earlier projects, you can pay attention to gitcon kernel, which can be understood as a common incubator in the Internet investment field. I am also a member of the first period of Gitcoin kernel. My intuitive feeling is that the information density under this information source is lower than the previous two, and the quality of developers is uneven.

Of course, Gitcoin is not a panacea. Andre, the founder of YFI, signed up for a certain period of Gitcoin Grand in order to obtain the support of code audit, but was overwhelmed by other projects. If you look at it from the logic of VC, it should be regarded as a huge directional step.

Some people describe Gitcoin as the "arsenal" of Ethereum. Every public chain hopes to have a Gitcoin, but only ETH has the community vitality to activate this product. The elements of this vitality include: 1. Rich and willing foundations 2. Active developers 3. Enthusiastic and slightly idealistic community users 4. The market required for project commercialization 5. Mature development experience

The authors of this article, Danilo Lessa Bernardineli, Michael Zargham and Jeff Emmett, use cadcad as a tool to analyze whether the Gitcoin allocation algorithm is effective in actual operation, and we will talk about the value of cadcad in the future.

In our previous analysis of the Gitcoin Grants ecosystem at BlockScience, we briefly introduced Quadratic Funding and the challenges this resource allocation policy faces in distributing matching funds fairly for community grants. In this article, we dive into how cadCAD can be used to identify specific instances of potential collusion or fraud, and discuss the consequences of policies that reduce exploitative behavior in the Gitcoin Grants ecosystem.

• suppose

• Throughout the Gitcoin Grants Round 8 network, grants are marked as blue nodes, contributors are marked as orange nodes, and donations serve as connections between them.

• Before starting...

• As a research institute dedicated to #OpenScience, we intend to approach the data analysis of the Gitcoin Grants ecosystem in a rigorous manner and present the results to an audience as we work. As such, you'll notice that this post takes the following format:

• assumed

• in conclusion

Methodology: Data and Models

explain and discuss

suppose

in conclusion

• Another important thing to be sure of before diving in, is the definition of "collusion" and "fraud", as these are the main concerns of the Gitcoin team in ensuring a fair and transparent public goods funding platform. According to Vitalik, who wrote an academic paper on CLR in Weyl & Hitzig, collusion is defined as "multiple agents for their mutual benefit to the detriment of other participants" and fraud is defined as "a citizen who misrepresents own identity". Analyzing these definitions is a bit tricky, because the saturation effect of the grants cycle actually implies that almost all community-level mobilization in the Gitcoin Grants system is collusion, since any gains in one community will harm the interests of other communities when sharing a finite pool . Also, Wikipedia defines collusion as "secret or unlawful cooperation or conspiracy, especially for the sole purpose of defrauding or defrauding others", which is more beneficial for our purposes because it allows us to explain intent when analyzing.

• Through this article, we will begin to examine the patterns of collaboration exhibited by the various communities participating in Gitcoin Grants, to see if we can identify collusion or fraudulent behavior, and how to mitigate it.

• suppose

There is an underlying "shape" for a community using Gitcoin Grants. Subjectively, the following formats can be used:

Dense communities, such as the Commons Stack / Token Engineering cluster, where there are a large number of related projects involving overlapping contributors.

Highly integrated communities, such as the Ethereum Foundation, where a large number of connections can be achieved without a tight cluster.

• Isolated communities, such as DAppChaser and other regional projects, where there are large numbers of project contributors and limited connections to other grants.

• Our hypothesis in this paper is that we can interpret these shapes as a kind of "fingerprint" for pattern matching organic community behavior, thereby better identifying predatory behavior and reducing the ability to attract matching funds from the entire Gitcoin community. But before we can determine the network structure of the colluding attack, we first have to understand the structure of the Gitcoin community. This shape can be detected quantitatively by using network science techniques such as community detection algorithms, which we explore further in the next section.

• assumed:

• In our analysis, we used several different tools and assumptions, which we list here due to the subjectivity they bring to our analysis.

Currency Aggregation This analysis condenses all donations into their dollar equivalent, regardless of the currency of the donation. This is a simplifying assumption of how the Gitcoin fund matching policy works. This is a simplifying assumption, following how Gitcoin's fund matching policy works.

Matching Funds Excluded This report does not include a rigorous analysis of matching funds computed through the Gitcoin Quadra Funding Algorithm, but instead focuses on the structure of the Gitcoin Grants network. Subsequent analysis can provide insight into the effects of the matching algorithm.

Community Detection Algorithm: Floating Communities Floating Communities is a propagation-based algorithm capable of identifying a variable number of communities in a dynamic network. They are based on fluids interacting in the environment, expanding and contracting due to this interaction. Liquid communities are able to find communities in synthetic graphs by exploiting accuracy close to the current best alternative. In this paper, we refer to these community swaps as "subgraphs".

Choice of community cut size: 5 We chose to divide it into 5 sub-communities: you always need to start somewhere, and 5 sub-communities is enough to give you an overview of the big picture, yet enough to make the structure understandable. Remember that the 5 subgraphs are cross-linked, so they can be pasted together to form the complete contribution graph.

Although we only chose a community detection algorithm and the number of subgraphs to split for a quick exploratory analysis, our research now aims to perform more rigorous sensitivity checks on hyperparameters such as the community clip size, via A range of measures such as modularity maximization, domain-specific supervised learning and percolation analysis of clustering coefficients are used to select algorithms. These choices will be revisited in the future to determine whether changes in algorithms or parameters could affect our conclusions and be consistent with reflective practice.

This is the entire Gitcoin Grants Round 8 network, this time represented by 5 separate community subgraphs, which we will explore further below.

Methodology: Data and Models

For this analysis, we are using Gitcoin Grants Round 8 donation data (before matching, as this does not affect the shape of the Gitcoin Grants network structure). We will input this data into the Gitcoin cadCAD model to perform our data analysis. With this model, we were able to estimate the eventual outcome of the game until any interventions by the Gitcoin team to mitigate the disruption were effective.

Using the cadCAD model, we generate a diagram representing the interconnections between grants and donors in the Gitcoin grant-giving network. To do this, we apply the floating community algorithm to a NetworkX graph object, which returns a list of detected communities. These communities consist of two nodes, grants and contributors, as nodes, with contributions flowing between them as edges.

We then visualize the community subgraph associated with three different authorizations: cadCAD subgraph, MetaGame subgraph, and DAppChaser subgraph.

A visualization of the Gitcoin Grant cooperation subgraph is shown below. To interpret the graphs, grants are represented as blue nodes and contributors as orange nodes. Grant nodes are sized by the total contributions of all donors, while donor (bonus) nodes are sized by their contributions to all grants. Edges between nodes represent donations flowing from contributors to grants.

Digging deeper into the data, we chose to analyze three (many) different community “fingerprints” to better understand the interaction patterns of different groups and how we might identify collusion or unfair behavior in Gitcoin Grants. cadCAD and Metagame were chosen to focus our analysis on communities with whom we had first-hand experience to improve our ability to interpret data in a social context. DappChaser was chosen to investigate the interesting patterns that emerge from this analysis.

Subgraphs of cadCAD

This subplot has an organic look, with a tight cluster of collaborations around the ecosystem of cadcad-related grants and ties to multiple projects within that ecosystem.

• image description

• Submap of neighboring communities for cadCAD grants, with grants in blue and contributors in orange.

• Wallkanda：2,Some statistics about the cadCAD subgraph: Total subgraph grants: 110 Total subgraph collaborators: 888 Total USD value for this subgraph: 41,047.76 USD

• Autonio：2,*Top five funding projects in this subgraph:**

• DistributedTown：2,Gitcoin Open Source Support Fund: $10,333.02

Commons Stack Community Fund — Panvala League: $2,628.53

$510.39

$500.89

$500.89

• Scribble Specifications and Runtime Verification Fund: 2,MetaGame Subgraph

• BeyondNFT: 2,This subgraph also exhibits an organic structure, albeit more diffuse, with lots of collaborative connectivity, but no dense clusters.

• Rekt- The dark web of DeFi journalism: 2,Subgraph of neighboring communities for Metagame grants, with grants in blue and contributors in orange

• Unitimes-Ethereum developer community platform: 1,Some statistics about the Metagame subgraph: Total funding: 171 Total collaborators: 825 Total dollar value in subgraph: $14,859.97 Top five funded projects in this subgraph:

• vfat.tools Yield Farming Calculators: 1,$958.19

$616.18

$552.87

$411.17

DAppChaser Subgraph

• KERNEL Block II [Panvala League]: 11,This network neighborhood looks a bit different compared to the other subgraphs. The network does not look as organic as our previous example, and there is little connection between donors and other Gitcoin grants in this ecosystem, suggesting less ecosystem collaboration.

• EU Crypto Initiative — educating policy makers: 9,image description

• Catnip: 3,Subgraph of neighboring communities for DappChaser grants, with grants in blue and contributors in orange.

• Meme of the Day: 2,Some stats about the DappChaser subgraph: Total Grants: 85 Total Collaborators: 966 Total Dollar Value in Subgraph: $26,243.74 Top 5 funded projects in this subgraph:

• COVID-19 Moment || Food Bag Delivery for families: 2,$755.68

$866.64

$038.81

$516.04

$118.16

With BrightID (identification system), these distinctions can be made more clearly. BrightID is an identity proof system that provides rewards for matching funds when collaborators are uniquely identified. The BrightID team is to be commended for their work on this particular model, as well as their efforts in identifying unique identities associated with grants. Given that Gitcoin created a grants pool specifically for projects in East Asia, it is clear that this is a systemic goal to attract more users from these regions. So while we need to be wary of collusion, we also need to ensure that any detection algorithms are not tuned too aggressively, preventing newcomers from different markets from joining future Gitcoin Grants rounds.

in conclusion

explain and discuss

In the analysis above, we can see certain patterns emerging in the data that can give us insight where to dig for more information. In DAppChaser cluster mode, an algorithm can flag potential collusion, but this problem may not be solved by only one algorithm. It may be necessary to build data collection and community governance tools designed to determine how policy is developed around collusion and how collusion policy is interpreted on a grant-by-grant basis when disputes arise about collusion. Ultimately, we are interested in how the Gitcoin system can elevate data about these actions to the governance layer of the system to resolve community disputes.

While this fingerprint can serve as a pattern for potential collusive behavior, we need to be careful that our algorithm is not naive enough to confuse a new entrant with a group of colluding actors, leading to unfair penalties. If our algorithm is too aggressive, we run the risk of shutting out new market entrants, which is also not good for the long-term growth of the Grants ecosystem.

• in conclusion

• Our goal in conducting this analysis is not to distinguish between "right" and "wrong" policy choices in the Gitcoin ecosystem, but to use data-driven analysis to explore subjective choices in algorithmic policy design. In the above discussion, we can clearly see that there is no objective "correct policy" that can solve all challenges in Gitcoin Grant matching. Therefore, a subjective choice must be made as to what compromise should be made to maintain the fairness of Gitcoin, given its community's norms and usage, and their desire for progressive decentralization. This analysis aims to provide clarity on these tradeoffs and provide decision support for the Gitcoin team and community in making policy choices to achieve their goals.

• We applaud Gitcoin for using a flagging algorithm that is able to identify areas of potential collusive behavior to appropriately direct the team's attention, and we are interested in how this algorithm can be further optimized. Given that "collusion" plays a larger role in Gitcoin Grants than discrete events, it is recommended to keep humans involved in collusion analysis and decision making.

While there are few hard answers when dealing with subjective measures like "fairness", there are some very good questions we can ask to help us gain insight into how exploitative we are in the Gitcoin Grants system so that we can policy accordingly choose: